2D Spiral

[Anonymous]

I am trying to create a 2D Spiral coil structure with trace spacings and trace width in order to mm. I have tried using sweep and extrude but they seem to work for less no of turns, not for turns exceeding 8 (No. of Turns > 8). Any help for trying to resolve this problem would be appreciated.

[leeminardi]

The failure of sweep may be more to do with the tightness of the curvature or the size of the shape to be swept than the number of turns.  Here is a sweep of a rectangle of 10 turns. 

Start with a top radius value that may be much greater than you want to see if you can get the number of turns you want. If successful try again with a smaller values until you reach a point where sweep doesn't work.  This will give you an indication of the threshold for the minimum radius at the start of the helix.

[Anonymous]

Hi Thank you for your reply, i did try to implement your solution but my structure is quite tight. The Base radius is 0.2 mm and top radius is 0.7 mm. Trace spacing is 0.01 mm and Trace Width is 0.01 mm. Your solution does not work. 

[WarrenGeissler]

Yeah @Anonymous - it depends on your number of turns vs. your radii. There is some strange cut-off algorithm that mathematically does not make sense. Anyone who wants to try it look at my drawing  (2013 format). The helix in the center is the only one that can be swept by the TINY circle at the start. So it's not the object intersecting itself. It has to be some strange formula of turns and radii, but for the life of me I cannot figure it out.

[SEANT61]

The underlying spline used to represent the parametric Helix entity may have a structure inconsistent with the requirements of the Sweep command.  Here, I've exploded the Helix (Spiral) to a spline entity, then reformatted it with a call to _CVREBUILD.  The resultant spline was much more uniform, and worked well enough with the Sweep command.

Not necessarily a straight forward process, but not too bad either. 

[WarrenGeissler]

CVREBUILD = SOOOPER Genius! I couldn't for the life of me get the resulting spline to work. Good one @SEANT61

[Anonymous]

Hi A Genius working with me came up with this idea:

[1] Make Helix -- > Specify Base Radius, Top Radius, No of Turns

[2] Use EXPLODE to break Helix into a SPLINE

[3] Use PEDIT --> YES --> Poly --> Acc. 10 --> Width (Specify 0.01 mm) 

Voila your done!!!!!

[Anonymous]

Hi can you please elaborate the steps: [1] Created Helix [2] Used 'EXPLODE' command [3] Used '_CVREBUILD' command.... after step [3] my helix turns into a line how do i reformat it? 

Your structure looks good but i want to know these steps in order to replicate and modify according to my expectations..

[leeminardi]

Rather than exploding a helix down to a spline and then to a polyline with small segmenta  I would prefer to keep a helix if possible.

The helix below was posted by @WarrenGeissler (I added the red line at 3 o'clock). The sweep command fails to work on this particular helix of the small circle along the helix. After adding the red line at a location in a few turns you can use the TRIM command to cut the helix into two pieces. The two pieces retain their helix identity. Sweep does work for these two pieces individually.  After sweeping you now have two solids which can be combined into one with union.  It may take a little experimentation to determine how far in you need to cut the helix in half to get results that can be swept.

Lee     

l

[WarrenGeissler]

@leeminardi - HOW on EARTH did you figure this out?? I would have never (In a million years) thought of breaking the spiral!

[leeminardi]

@WarrenGeissler My first guess at why sweep didn't work was because the instantaneous radius of curvature at the start of the helix was too small but your examples showed that there was no direct correlation between the radius of curvature and sweep not working.  It appeared that total helix length and curvature may have had an effect so I just lopped off a few turn and it worked! 

[elshawadfy]

Hi @Anonymous!

The sweep fails when the width of the swept shape exceeds the distance between tow consequent turns or lines, because the resulting sweep intersects with itself..

So when  you increase the number of turns to the point where the distance between turns is less than that width, you get the error message and the sweep fails..

You then need either to make the swept section smaller, or adjust your inner or outer radius, or number of turns / to increase the distance between turns..

[leeminardi]

@Anonymous"The sweep fails when the width of the swept shape exceeds the distance between tow consequent turns or lines, because the resulting sweep intersects with itself.."

This is not true.  Check out the drawing @WarrenGeissler posted.  A helix with a closer spacing worked whereas a larger spacing does not work.

My theory is that when the total number of CVs exceeds some threshold or the radius of curvature is too small then the sweep operation fails.  For the drawing cited, the failure is immediate indicating (potentially) that the calculation of the solid was never really started as some other issue was indicated.  

[SEANT61]

---

@Anonymous wrote:

Hi can you please elaborate the steps: [1] Created Helix [2] Used 'EXPLODE' command [3] Used '_CVREBUILD' command.... after step [3] my helix turns into a line how do i reformat it? 

Your structure looks good but i want to know these steps in order to replicate and modify according to my expectations..

---

The CRVREBUILD command will use pre-existing parameters.  Reset those to match that of the incoming parameters, and the command will make the distribution more uniform.

[elshawadfy]

Hi @leeminardi!

What I said about the cross section is true, AutoCAD will always refuse to do the sweep if the resulting geometry self intersects, be that because of the cross section is wider than the spacing between the path segments or any other reason.. [See Pic.]

What @WarrenGeissler file shows is that there are other cases beside the cross section being wider than the spacing, where AutoCAD considers the resulting geometry to be self intersecting.. 

There is a small problem also with with @SEANT61 is that the curve resulting from _CVREBUILD is very different from the original @ same number of control points, you nee to increase the Control Vertices alot to match the original shape although they seem the same.. - (The swep still succeeds, which indicates the issue has to do more with the CVs distribution more than their count)

Oddly enough, breaking the right and left spirals (in @WarrenGeissler file) into pieces using "break at one point" and then sweeping along the parts results in a success!!!! And the resulting parts accept Union!!! (And with break, the difference from the original spiral is barely noticeable)

What that may suggest in my opinion, is that this isn't just about the curvature, because that shouldn't change with break, (or should it? - not really sure).. Nor is it because of the CVs count, because increasing the CVs significantly after _CVREBUILD still resulted in a success with sweep.. 

Another odd thing: changing the swept shape for the first spiral to the left into a rectangle (even wider than the circle originally used) resulted in a success!!!

 

So the only conclusion I could come up with, that the problem with the spirals that failed has to do with the  relation between the vertices distribution of both the path and the cross section..

Regards!

[WarrenGeissler]

@elshawadfy,

Just - WOW. Your depth of study on this is amazing. 

[SEANT61]

---

There is a small problem also with with @SEANT61 is that the curve resulting from _CVREBUILD is very different from the original @ same number of control points, you nee to increase the Control Vertices alot to match the original shape although they seem the same..

---

This statement and the accompanying picture in you previous response does not match what I've encountered. 

In my linked file here,  I've set up the spiral to what I believe is the parameters listed by axm179431.  When exploded and the CRVREBUILD command is evoked, the incoming parameters are 334 CVs and Degree 3.  Reusing those parameters does not modify the curve any significant amount.

Can you post the file you used in your investigation?

[SEANT61]

As a matter of fact, your picture demonstrating a geometry change resulting from the CVREBUILD command  compares two different Helixes.  In the picture I see 10 different turns for Original and 500CV, but with the 192CV I see 12.

[elshawadfy]

Hi everyone!

Thank you very much @WarrenGeissler, hope we find out exactly why AutoCAD was refusing the sweep in these spirals..

@SEANT61, my sincere apologies, I experimented a lot with the spirals, I used the spiral to the left in @WarrenGeissler file, but it seems that it was  modified accidentally along the way. when I reapplied the _CVREBUILD this time it was very close, no such big differences.. Thank you for pointing out the number of turns, and I apologize again..

Regards, 

[SEANT61]

No problem. 

It's good that AutoCAD retains interest enough to get several users motivated to figure out one of it's quirks.